DMSC Working Group X. Li
Internet-Draft China Telecom
Intended status: Standards Track B. Liu
Expires: 21 January 2027 Huawei Technologies
J. Liu
Beijing University of Posts and Telecommunications
C. Du
Tsinghua
L. Zhang
AsiaInfo Technologies (China) Inc
20 July 2026
Multi-agent Collaboration Protocol Suites Architecture
draft-li-dmsc-macp-06
Abstract
This document defines a protocol suite and architectural framework
for secure and scalable multi-agent collaboration. The proposed
Multi-Agent Collaboration Protocol (MACP) enables trusted agent
onboarding, capability-based query, distributed capability
synchronization, and secure interaction among agents and external
resources. The architecture introduces key entities such as the
Agent Management Center (AMC), Agent Gateway (AGW), Agents, and
External Resource Services (ERS), along with a set of protocols that
collectively support dynamic, capability-driven collaboration across
administrative domains.
Status of This Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at https://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on 21 January 2027.
Li, et al. Expires 21 January 2027 [Page 1]
Internet-Draft MACP July 2026
Copyright Notice
Copyright (c) 2026 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents (https://trustee.ietf.org/
license-info) in effect on the date of publication of this document.
Please review these documents carefully, as they describe your rights
and restrictions with respect to this document. Code Components
extracted from this document must include Revised BSD License text as
described in Section 4.e of the Trust Legal Provisions and are
provided without warranty as described in the Revised BSD License.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Conventions used in this document . . . . . . . . . . . . . . 4
3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4
4. Multi-agent Collaboration Protocol Architecture . . . . . . . 5
4.1. Agent Management Center . . . . . . . . . . . . . . . . . 6
4.2. Agent Gateway . . . . . . . . . . . . . . . . . . . . . . 7
4.3. Agent . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4.4. External Resource Service (ERS) . . . . . . . . . . . . . 8
4.5. Data Objects . . . . . . . . . . . . . . . . . . . . . . 9
4.6. Entity Summary . . . . . . . . . . . . . . . . . . . . . 10
5. Multi-Agent Collaboration Protocol Suite Overview . . . . . . 10
5.1. Agent-Gateway Interaction Protocol(AGIP) . . . . . . . . 10
5.1.1. Agent Registration and Capability Advertisement . . . 11
5.1.2. Capability Resolution . . . . . . . . . . . . . . . . 12
5.2. Agent Authentication and Authorization Protocol (AAAP) . 12
5.3. Capability Directory Synchronization Protocol (CDSP) . . 13
5.4. Gateway-side Mediation Requirements . . . . . . . . . . . 15
5.5. Agent to External Resource Service Protocol . . . . . . . 16
6. Capability Model . . . . . . . . . . . . . . . . . . . . . . 16
7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 17
8. Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . 17
9. Normative References . . . . . . . . . . . . . . . . . . . . 17
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 18
1. Introduction
The rapid evolution of large-scale multi-agent systems introduces new
requirements for coordination, security, and service discovery across
distributed environments. Agents are no longer confined to isolated
execution contexts, but increasingly operate across administrative
domains, network boundaries, and heterogeneous infrastructures.
However, existing mechanisms for service interaction and discovery
Li, et al. Expires 21 January 2027 [Page 2]
Internet-Draft MACP July 2026
exhibit several limitations when applied to multi-agent
collaboration:
* Lack of unified trust establishment: Current agent interaction
models often assume pre-established trust or rely on application-
layer authentication, without a network-level mechanism to ensure
that participating agents are authenticated, authorized, and
accountable across domains.
* Insufficient capability abstraction and discoverability:
Traditional service discovery mechanisms (e.g., DNS-based or
registry-based approaches) focus on endpoint resolution rather
than capability-oriented matching, making them unsuitable for
dynamic agent collaboration where tasks are fulfilled based on
functional capabilities rather than fixed service locations.
* Limited visibility across distributed environments: Existing
systems lack a mechanism to construct a distributed, up-to-date
view of available agent capabilities, especially when agents are
registered under different control points or administrative
domains.
* Inefficient or ad hoc discovery mechanisms: Without coordinated
discovery strategies, agent systems rely on broadcast-like or
centralized queries, leading to scalability challenges and
increased latency in locating suitable collaborators.
* Fragmented protocol landscape: While protocols such as A2A, MCP,
or other interaction mechanisms exist, they operate in isolation
and do not provide an integrated framework for authentication,
registration, discovery, and coordination.
These limitations become more critical as multi-agent systems scale,
where dynamic task composition, cross-domain collaboration, and
secure interaction are fundamental requirements. To address these
challenges, this document proposes the Multi-Agent Collaboration
Protocol (MACP), a protocol suite and architectural framework that:
* Establishes a trusted onboarding mechanism via a centralized
authentication and authorization entity.
* Introduces capability-based abstraction and identification for
agents.
* Enables distributed capability synchronization across control
points.
Li, et al. Expires 21 January 2027 [Page 3]
Internet-Draft MACP July 2026
* Integrates existing interaction protocols into a cohesive
collaboration framework.
By shifting from endpoint-centric interaction to capability-driven
collaboration, MACP enables scalable, secure, and flexible multi-
agent systems that can operate effectively across heterogeneous and
distributed environments.
2. Conventions used in this document
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in [RFC2119].
3. Terminology
The following terms are defined in this draft:
* Agent: An automated intelligent entity capable of e.g interacting
with its environment, acquiring contextual information, reasoning,
self-learning, decision-making, executing tasks (autonomously or
in collaboration with other Al Agents) to achieve a specific goal.
* Agent Gateway: The Agent Gateway is a functional entity that
serves as the infrastructure for enabling interconnection and
collaboration among agents. While its core role remains
consistent, it is inherently flexible in deployment and can be
realized in various forms—ranging from a network service to a
dedicated gateway—depending on the architectural and operational
requirements of different network environments.
* Agent Management Center (AMC): It is the trusted infrastructure
service responsible for agent identity lifecycle management and
credential issuance.
* Agent Identity Code (AIC): An Agent Identity Code (AIC) is a
verifiable, globally unique identifier that represents the
identity of an Agent.
* Agent Capability Specification (ACS): An Agent Capability
Specification (ACS) is a structured description of an agent's
capabilities and service information that can be stored,
retrieved, and matched.
Li, et al. Expires 21 January 2027 [Page 4]
Internet-Draft MACP July 2026
* Agent Credential: An Agent Credential is a tamper-resistant data
object issued by an Agent Management Center(or its credential
authority component), used by an Agent to prove identity
attributes and/or authorization to a relying party. Examples
include X.509 certificates and security tokens.
* Agent-Gateway Interaction Protocol (AGIP): The AGIP defines the
control-plane interaction between an Agent and its locally
attached Agent Gateway. It provides mechanisms for agent
onboarding, capability registration, lifecycle management, and
capability resolution, enabling an Agent to obtain suitable
collaboration peers based on capability requirements.
* Agent Authentication and Authorization Protocol (AAAP): The AAAP
defines how authentication and authorization decisions are
requested and enforced.
* Capability Directory Synchronization Protocol (CDSP): The CDSP
synchronizes abstracted agent capability digests across agent
gateways.
4. Multi-agent Collaboration Protocol Architecture
The MACP architecture consists of the following key entities:, as
shown in figure 1. Each functional entity represents a logical role
in the IoA architecture, implementations MAY combine multiple
entities into a single product.
Li, et al. Expires 21 January 2027 [Page 5]
Internet-Draft MACP July 2026
+-----------------------+
|Agent Management Center|
| (Authentication & |
| Authorization) |
+-----------------------+
| AAAP | AAAP
| |
+-------------------+ +--------------------+
| Agent Gateway 1 | | Agent Gateway 2 |
| + Registration |----------| + Registration |
| + Capa Resolution | CDSP | + Capa Resolution |
| + Capa Sync | | + Capa Sync |
+--+-----+----------+ +--------+-----+-----+
| |
AGIP | AGIP |
| |
+-----------------------+ +----------------------+ +-----------------+
| Agent | A2A | Agent | MCP |External Resource|
| (Role A) |----------- | (Role B) |------- | Service (ERS) |
+----------+------------+ +----------------------+ +-----------------+
| |
+----------+------------+ +----------+------------+
| User A | | User B |
+-----------------------+ +-----------------------+
Figure 1 MACP Architecture Overview
4.1. Agent Management Center
The Agent Management Center is the trusted infrastructure service
responsible for agent identity lifecycle management and credential
issuance. The AMC provides centralized authentication and
authorization services. It ensures that only legitimate and trusted
agents are allowed to join the system. Specifically, the AMC:
* Authenticates agent identity
* Determines authorization scope and execution permissions
* Issues authorization credentials (e.g., certificates or tokens)
Multiple Agent Management Center MAY exist in an IoA deployment, each
managing a subset of agents within its administrative scope. A
deployment MAY realize the identity management function and the
credential authority function as separate services, provided they
maintain consistent identity-to-credential binding.
Li, et al. Expires 21 January 2027 [Page 6]
Internet-Draft MACP July 2026
4.2. Agent Gateway
The Agent Gateway is a functional entity that serves as the
infrastructure for enabling interconnection and collaboration among
agents. While its core role remains consistent, it is inherently
flexible in deployment and can be realized in various forms—ranging
from a network service to a dedicated gateway—depending on the
architectural and operational requirements of different network
environments.
The Agent Gateway provides the following functions:
* Agent Registration: Maintains agent identity and capability
information.
* Agent Capability Directory Management: Stores and organizes
registered agent capabilities.
* Agent Capability Directory Synchronization: Exchanges agent
capability digests with peer gateways.
* Agent Group Communication Support: Enables multi-agent
coordination.
Each AGW maintains a local capability view and participates in
forming a distributed capability knowledge plane. More specific
requirements are specified in [draft-liu-dmsc-gw-
requirements][GW-REQ].
4.3. Agent
The Agent is an automated intelligent entity capable of e.g
interacting with its environment, acquiring contextual information,
reasoning, self-learning, decision-making, executing tasks
(autonomously or in collaboration with other Al Agents) to achieve a
specific goal.
An Agent is responsible for:
* Maintaining its own identity information (e.g., AIC) and
credentials locally.
* Maintaining its capability description (e.g., ACS) and ensuring
consistency with its current state.
Li, et al. Expires 21 January 2027 [Page 7]
Internet-Draft MACP July 2026
* Performing authentication and authorization checks for
interconnection, including mutual verification of peer agents and
validation of presented credentials. - Conducting agent-to-agent
interaction, including session establishment, message exchange,
and task/context management.
* Accessing external resources when required to fulfill tasks.
* Producing monitoring and logging data for troubleshooting,
auditing, and governance purposes.
These are internal agent capabilities described here for
informational purposes. They are NOT standardized as separate
architectural functional components. In an interaction, an Agent MAY
assume different roles depending on the collaboration mode. The DMSC
architecture does not constrain the set of possible roles; specific
collaboration protocols MAY define role semantics appropriate to
their interaction patterns.
For example, in a task-driven collaboration [draft-yang-dmsc-ioa-
task-protocol] [draft-yang-dmsc-ioa-task-protocol]:
* Leader: The Agent that initiates tasks and organizes
collaboration.
* Partner: The Agent that accepts tasks and provides services,
executing assigned tasks and returning results to the Leader.
A single Agent implementation MAY act in different roles across
different interactions. Role assignment is per-interaction, not per-
deployment.
4.4. External Resource Service (ERS)
The External Resource Service (ERS) represents external systems such
as APIs, databases, or compute services that agents may invoke. ERS
is conceptually external to the agent collaboration system and is
accessed via existing protocols.
ERS may include both domain-specific services and shared
infrastructure services. In particular, certain ERS instances MAY
correspond to widely deployed Internet-scale infrastructure (e.g.,
naming, data access, or knowledge retrieval systems), which provide
common capabilities that are not specific to agent collaboration but
are essential for its operation. In this sense, ERS can be viewed as
leveraging existing or future shared Internet service infrastructure,
rather than replicating such functionality within the agent system
itself.
Li, et al. Expires 21 January 2027 [Page 8]
Internet-Draft MACP July 2026
A key design principle is the separation of responsibilities between
the agent collaboration system and ERS. The agent system is
responsible for:
* Agent identity, trust establishment, and authorization
* Agent capabilities registration, abstraction
* Coordination and interaction among agents
In contrast, ERS is responsible for:
* Providing external data, computation, or domain-specific
functionality
* Supporting tasks that require capabilities beyond the agent system
itself
Agents interact with ERS when executing tasks that require external
resources, while core collaboration functions—such as routing, and
coordination—remain within the agent system. The MACP architecture
intentionally avoids redefining general-purpose Internet services
(e.g., naming or data retrieval), and instead focuses on enabling
agents to discover, select, and utilize such services in a
coordinated manner.
4.5. Data Objects
The following are protocol data objects referenced by the functional
entities. They are not functional entities themselves:
* Agent Identity Code (AIC): An Agent Identity Code (AIC) is a
verifiable, globally unique identifier that represents the
identity of an Agent. An AIC is allocated by an Agent Gateway
during agent registration.
* Agent Capability Specification (ACS): An Agent Capability
Specification (ACS) is a structured description of an agent's
capabilities and service information that can be stored,
retrieved, and matched. An ACS MAY use the JSON [RFC8259]
[RFC8259] format, typically including: the agent's AIC, functional
capabilities, technical characteristics, service interfaces, and
security requirements.
Li, et al. Expires 21 January 2027 [Page 9]
Internet-Draft MACP July 2026
* Agent Credential: An Agent Credential is a tamper-resistant data
object issued by an AMC (or its credential authority component),
used by an Agent to prove identity attributes and/or authorization
to a relying party. Examples include X.509 certificates and
security tokens.
4.6. Entity Summary
The following table provides a summary of all functional entities and
external actors in the simplified DMSC architecture.
| # | Entity | Type | Role in Architecture |
|----|---------------------------|-------------------|-------------------------------------------------------|
| 1 | Agent | Core entity | Autonomous task execution and collaboration |
| 2 | Agent Management Center | Infrastructure | Identity lifecycle management and credential issuance |
| 3 | Agent Gateway | Infrastructure | Capability directory, synchronization |
| 4 | External Resource Service | Infrastructure | External resource exposure and invocation handling |
| 5 | User | External actor | Task initiation, authorization, and result consumption|
Figure 2 A summary of all functional entities
5. Multi-Agent Collaboration Protocol Suite Overview
MACP defines a set of protocols to enable interaction among entities.
5.1. Agent-Gateway Interaction Protocol(AGIP)
The Agent-Gateway Interaction Protocol (AGIP) defines the control-
plane interaction between an Agent and its locally attached Agent
Gateway (AGW). AGIP provides a unified interaction framework that
enables Agents to join the collaboration system, advertise and
maintain their capabilities, and request capability resolution
services throughout their lifecycle.[draft-sz-dmsc-iaip]
As the primary interface between an Agent and the AGW, AGIP supports
two complementary interaction procedures. The first enables an Agent
to establish its identity and advertise its capabilities to the
gateway, allowing the AGW to maintain an up-to-date capability
directory of locally attached Agents. The second enables an Agent to
submit collaboration requests during runtime. Upon receiving such
requests, the AGW interprets the requested capabilities, resolves
them against its distributed capability knowledge, and returns one or
more suitable collaboration Agents.
Accordingly, AGIP consists of the following two procedures: Agent
Registration and Capability Advertisement, Capability Resolution.
Li, et al. Expires 21 January 2027 [Page 10]
Internet-Draft MACP July 2026
5.1.1. Agent Registration and Capability Advertisement
This procedure enables an Agent to become a trusted participant in
the collaboration system while publishing its capabilities to the
attached Agent Gateway.
AGIP is responsible for:
* Agent identity declaration.
* Capability advertisement.
* Agent registration and deregistration.
* Capability update and lifecycle management.
Operational flow:
1. The Agent sends a registration request to its attached AGW,
carrying its identity information together with its Agent
Capability Specification (ACS).
2. The AGW invokes the Agent Authentication and Authorization
Protocol (AAAP), as described in Section 5.2, to authenticate and
authorize the requesting Agent.
3. Upon successful authentication and authorization, the AGW
allocates a globally unique Agent Identity Code (AIC).
4. The AGW stores the advertised capability information in its local
capability directory and associates it with the allocated AIC.
5. The AGW acknowledges successful registration to the Agent.
6. The registered capability information becomes available for
subsequent synchronization with peer AGWs through the Capability
Directory Synchronization Protocol (CDSP).
After registration, an Agent MAY update its capability information
whenever its supported functions, service interfaces, execution
status, or other capability-related attributes change. Likewise, an
Agent MAY deregister from the AGW when it leaves the collaboration
system, allowing the corresponding capability information to be
withdrawn from the capability directory.
Li, et al. Expires 21 January 2027 [Page 11]
Internet-Draft MACP July 2026
5.1.2. Capability Resolution
After successful registration, an Agent MAY request collaboration
services by submitting a capability request to its attached AGW
through AGIP.
Upon receiving a request from an agent, the AGW performs the
following operations:
1. Receive the capability request from the requesting Agent.
2. Interpret and normalize the requested capability into an internal
capability representation.
3. Match the normalized capability against the capability directory
maintained by the AGW.
4. Identify one or more candidate Agents capable of satisfying the
requested capability.
5. Apply local selection policies when multiple candidate Agents are
available.
6. Return the capability resolution result to the requesting Agent.
The capability directory maintained by the AGW contains both
capabilities advertised by locally registered Agents and abstract
capability summaries synchronized from peer AGWs through CDSP.
Consequently, capability resolution may identify candidate Agents
either within the local administrative domain or in remote domains
without requiring the requesting Agent to interact directly with
multiple gateways.
When multiple candidate Agents satisfy the requested capability, the
AGW MAY rank or select candidates according to locally configured
policies, including capability matching degree, resource
availability, execution load, trust relationships, network
conditions, administrative policies, or application-specific
optimization objectives.
5.2. Agent Authentication and Authorization Protocol (AAAP)
The Agent Authentication and Authorization Protocol (AAAP) is used
between the Agent Gateway (AGW) and the Agent Management Center (AMC)
to establish trust for agents attempting to join the system.
AAAP is responsible for:
Li, et al. Expires 21 January 2027 [Page 12]
Internet-Draft MACP July 2026
* Verifying the identity of an agent (via AGW as a proxy)
* Determining the authorization scope and permitted operations
* Issuing authorization credentials (e.g., tokens or certificates)
Operational flow:
1. The AGW receives a registration request from an agent.
2. The AGW initiates an AAAP request to the AMC, carrying agent
identity information.
3. The AMC performs authentication and authorization checks.
4. Upon success, the AMC returns an authorization credential and
policy constraints.
5. The AGW enforces the received authorization decision.
6. The AMC acts as the trust anchor of the system, while the AGW
acts as the policy enforcement point (PEP).
Note that although AAAP establishes the initial trust relationship
between an agent and the system (i.e., onboarding trust), subsequent
interactions between agents may require additional, context-specific
authentication and authorization. Such interaction-level mechanisms
are out of scope for AAAP and MAY leverage existing frameworks (e.g.,
OAuth-based token exchange or similar delegation mechanisms) to
support secure, fine-grained access control between agents.
An agent MUST successfully complete AAAP before participating in
invocation or collaboration. However, successful onboarding via AAAP
does not eliminate the need for authentication and authorization
during runtime interactions between agents.
5.3. Capability Directory Synchronization Protocol (CDSP)
The Capability Directory Synchronization Protocol (CDSP) is used
between Agent Gateways (AGWs) to synchronize capability directory
information and construct a distributed view of agent capabilities
across the network. More detailed treatment of the gateway
capability directory in [draft-zhang-dmsc-gateway-directory-sync]
[draft-zhang-dmsc-gateway-directory-sync].
CDSP is designed to:
Li, et al. Expires 21 January 2027 [Page 13]
Internet-Draft MACP July 2026
* Enable distributed visibility of agent capabilities across
multiple AGWs
* Avoid centralized bottlenecks in capability management
* Preserve privacy and scalability through abstraction
* Support incremental and policy-controlled synchronization.
Each AGW maintains a local capability directory, which is populated
through agent registration and updated over time. CDSP enables AGWs
to exchange selected portions of these directories so that capability
information is not confined to a single gateway but becomes visible,
in an abstracted form, across multiple administrative or network
domains. To ensure scalability and protect sensitive information,
CDSP does not transfer complete capability descriptions. Instead,
AGWs exchange capability directory entries in a summarized form.
Each entry represents a capability exposed by an agent and is
associated its corresponding capability vector. The exchanged
information MAY include semantic descriptions or structured
representations of the capability, along with limited metadata such
as version or category. Detailed implementation-specific information
and sensitive attributes MUST NOT be propagated through CDSP.
CDSP supports different synchronization scopes depending on
deployment requirements. During initial establishment between peer
AGWs or recovery scenarios, a gateway MAY perform a full
synchronization of its capability directory. In steady-state
operation, synchronization is typically incremental or selective,
where only updated or policy-permitted entries are exchanged. The
selection of entries MAY be governed by administrative policies,
trust relationships, or capability classification. Synchronization
can be triggered in multiple ways. An AGW MAY initiate periodic
synchronization to maintain freshness of the distributed view. It
MAY also perform event-driven updates when local changes occur, such
as agent registration, deregistration, or capability updates.
Additionally, synchronization MAY be requested on demand by peer
gateways when needed for coordination purposes.
Given the distributed nature of CDSP, strict consistency across all
AGWs is not required. Instead, the system operates under an eventual
consistency model, where capability directory views converge over
time. To support this, capability entries SHOULD include versioning
or timestamp information, allowing AGWs to reconcile updates and
prefer the most recent information. Conflict resolution policies MAY
be applied when inconsistencies arise. All CDSP exchanges MUST be
authenticated and integrity-protected.
Li, et al. Expires 21 January 2027 [Page 14]
Internet-Draft MACP July 2026
5.4. Gateway-side Mediation Requirements
The protocols defined in this document specify how Agents, Agent
Gateways (AGWs), and gateway support functions exchange registration
information, capability information, authentication information, and
collaboration requests. These protocols provide the information
required for gateway-assisted collaboration. However, they do not
specify how an Agent Gateway transforms the exchanged information
into an operational collaboration decision.
In practical deployments, an Agent Gateway performs more than
protocol forwarding, capability directory management, or message
relay. When an Agent submits a collaboration request through the
Agent-Gateway Interaction Protocol (AGIP), the gateway is responsible
for determining whether the request can be satisfied, which Agent or
service is appropriate, and under what operational constraints the
collaboration should proceed. Although the decision logic may vary
across deployments, the need for gateway-side decision making is
common across heterogeneous agent ecosystems.
To produce an appropriate collaboration decision, the gateway may
need to interpret the request objective, correlate it with registered
capability descriptions, evaluate candidate capabilities, apply
authorization and operational policies, consider trust relationships,
and assess execution context before selecting an appropriate
collaboration target. These decision inputs are derived from the
protocol interactions defined in MACP but require additional gateway-
side processing before they can be used for collaboration decisions.
The decision process may also incorporate information obtained from
multiple gateway support functions. Capability directories
synchronized through CDSP provide candidate capability views.
Authentication and authorization services defined by AAAP provide
identity and permission information. Registration information
maintained through AGIP provides capability metadata and lifecycle
state. Operational policies, administrative constraints, network
conditions, and deployment-specific knowledge may further influence
the gateway's decision. These information sources collectively
contribute to gateway-side decision making but do not themselves
define how decisions are produced.
After a collaboration decision has been made, the gateway may further
determine how the request should be handed over to the selected
interaction mechanism. Depending on deployment requirements, the
gateway may prepare additional information describing the selected
capability, applicable policy constraints, interaction context,
routing information, trust context, or failure reasons. The
resulting decision and handoff context enable different interaction
Li, et al. Expires 21 January 2027 [Page 15]
Internet-Draft MACP July 2026
mechanisms, including A2A, MCP, IAIP, or other protocols, to execute
collaboration while preserving consistent capability, policy, and
trust semantics.
The gateway-side decision process described above complements the
protocol interactions defined by MACP. While MACP specifies how
collaboration information is exchanged among Agents and Agent
Gateways, it does not define how gateways interpret requests or
derive collaboration decisions from the exchanged information.
This motivates the need for a gateway mediation function. A
companion draft, Gateway Mediation Layer
[draft-zhang-dmsc-ioa-semantic-interaction], analyzes the mediation
inputs, decision process, outputs, and interoperability
considerations required for gateway-assisted collaboration across
heterogeneous agent environments.
5.5. Agent to External Resource Service Protocol
Interaction between agents and external resources (ERS) is supported
via existing protocols such as:
* Model Context Protocol (MCP)
* Agent-to-Tool (A2T)
MACP does not redefine these protocols but enables their integration
within the architecture.
6. Capability Model
Capabilities are the core abstraction in MACP.
* Each capability is associated with a Capability vector.
* Capabilities are registered, indexed, and discovered via AGWs.
* Capability information is abstracted during synchronization to
protect sensitive details.
* Capability descriptions MAY be semantic or structured depending on
use case.
This abstraction enables flexible and scalable service composition.
Li, et al. Expires 21 January 2027 [Page 16]
Internet-Draft MACP July 2026
7. IANA Considerations
TBD
8. Acknowledgement
TBD
9. Normative References
[draft-sz-dmsc-iaip]
S, S., "Intent-based Agent Interconnection Protocol at
Agent Gateway. draft-sz-dmsc-iaip.
", 9
February 2026.
[draft-verma-dmsc-nlip-notes]
V, D., "Use of Natural Language for Agent Communication.
draft-verma-dmsc-nlip-notes.
", 11 February 2026.
[draft-yang-dmsc-ioa-task-protocol]
Y, C., "Internet of Agents Task Protocol (IoA Task
Protocol) for Heterogeneous Agent Collaboration. draft-
yang-dmsc-ioa-task-protocol.
", 14 January 2026.
[draft-zhang-dmsc-gateway-directory-sync]
Z, L., "Gateway Capability Directory and Synchronization
for Internet of Agents. draft-zhang-dmsc-gateway-
directory-sync. ", 28 April 2026.
[draft-zhang-dmsc-ioa-semantic-interaction]
Z, L., "Ontology-based Semantic Interaction for Internet
of Agents. draft-zhang-dmsc-ioa-semantic-interaction.
", 4 February 2026.
[GW-REQ] L, B., "Gateway Requirements for Dynamic Multi-agents
Secured Collaboration. draft-liu-dmsc-gw-requirements.
", 16 January 2026.
Li, et al. Expires 21 January 2027 [Page 17]
Internet-Draft MACP July 2026
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119,
DOI 10.17487/RFC2119, March 1997,
.
[RFC8259] Bray, T., Ed., "The JavaScript Object Notation (JSON) Data
Interchange Format", STD 90, RFC 8259,
DOI 10.17487/RFC8259, December 2017,
.
Authors' Addresses
Xueting Li
China Telecom
Beiqijia Town, Changping District
Beijing
Beijing, 102209
China
Email: lixt2@foxmail.com
Bing Liu
Huawei Technologies
No. 156 Beiqing Road
Beijing
China
Email: leo.liubing@huawei.com
Jun Liu
Beijing University of Posts and Telecommunications
10 Xitucheng Road, Haidian District
Beijing
100876
China
Email: liujun@bupt.edu.cn
Chenguang Du
Tsinghua
Beijing
100094
China
Email: du_chenguang@outlook.com
Li, et al. Expires 21 January 2027 [Page 18]
Internet-Draft MACP July 2026
Lianhua Zhang
AsiaInfo Technologies (China) Inc
Beijing
100000
China
Email: zhanglh2@asiainfo.com
Li, et al. Expires 21 January 2027 [Page 19]